[Beowulf] Not quite Walmart, or, living without ECC?

[William Dinkel]

On Nov 27, 2007, at 1:52 PM, David Mathog wrote:

> Michael Will wrote:
>
>> We have found that linpack is by far the better memory tester than
>> Memtest86+.
>
> So now we have a report of a second method that finds more memory
> problems than memtest86+.  Can somebody please shed some light on why
> these two programs find defects in memory that memtest86+ doesn't?
> Or is it that they find defects in other parts of the hardware,
> external to the actual RAM, which manifest as memory errors?  The
> key distinction being that swapping memory sticks will cure the
> former but not the latter.
>
> In any case I'd like to know what it is about linpack/memtester which
> lets them find memory faults that memtest86+ doesn't.  Presumably
> whatever this magic sauce is could be added to memtest86+, once
> again resulting in a tiny memory tester which can run without the rest
> of linux.  That is a desirable goal, since the rest of linux sits on a
> largish chunk of memory which cannot be tested with either linpack or
> memtester, and which is critical to system stability.

Heat?  Just a quick guess, but in the linpack case the entire system  
is generally warmer than during a simple memtest86+ run.

William Dinkel
Chief Technology Officer
Team HPC
http://www.teamhpc.com
wdinkel at teamhpc.com
1-866-TEAMHPC



-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://www.beowulf.org/pipermail/beowulf/attachments/20071127/c7c54ea1/attachment.html>